https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License.
1 Global Distribution and 14-YearRadiation Changes amplif in Erythemal Irradiance, UV Atmospheric Transmission, and Total Column Ozone 2005 – 2018 Estimated from OMI and EPIC 2 Observations 3
4 Jay Herman1 Alexander Cede2 Liang Huang3 Jerald Ziemke5 5 Matthew Kowalewski2, Karin Blank4 6
7
1University of Maryland Baltimore County JCET, Baltimore, Maryland USA
2SciGlob Instruments and Services, Ellicott City, Maryland, USA
3Science Systems and Applications, Lanham, Maryland, USA
4NASA Goddard Space Flight Center, Greenbelt, Maryland USA
5Morgan State University, GESTAR, Baltimore Maryland
Corresponding Author email: [email protected]
1 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License.
8 Abstract
9 Satellite data from the Ozone Measuring Instrument (OMI) and Earth Polychromatic Imaging Camera 10 (EPIC) for ozone amount and scene reflectivity (mostly from clouds) are used to study changes and global 11 distribution of UV erythemal irradiance in mW/m2 E(,z,t) and UV index (E/25 mWm2) over the Earth’s 12 surface as a function of latitude , longitude altitude z, and time t. OMI time series data starting in
13 January 2005 to December 2018 are used to estimate 14-year changes in total column ozone TCO3 and 14 scene reflectivity at 105 specific land plus 77 ocean locations in the Northern and Southern Hemispheres. 15 Estimates of changes in atmospheric transmission T(,z,t) derived from cloud and haze reflectivity show 16 almost no average 14-year change from 55OS to 35ON but show an increase from 40ON to 60ON. This 17 implies increased solar insolation at high northern latitudes that suggests positive feedback for global O O 18 warming. TCO3 has increased at a rate of 2% per decade for the latitudes between 60 S to 10 N changing 19 to a decrease of 1% per decade between 40ON to 60ON. The result is an average decrease in E(,z,t) at a 20 rate of 2% per decade in the Southern Hemisphere and an increase between 40ON to 60ON. For some 21 specific sites (latitudes from 55OS to 45ON) there has been little or no change in E(,z,t) for the period 22 2005 – 2018. Nearly half the sites show the effects of both short- and long-term cloud change as well as 23 total column ozone change. Synoptic EPIC data from the sunlit Earth are used to derive ozone and 24 reflectivity needed for global images of the distribution of E(,z,t) from sunrise to sunset centered on 25 the Americas, Europe-Africa, and Asia. EPIC data are used to show the latitudinal distribution of E(,z,t) 26 from the equator to 75O for specific longitudes. Dangerously high amounts of erythemal irradiance (12 < 27 UV index < 18) are found for many low latitude and high-altitude sites (e.g., San Pedro, Chile (2.45 km), La 28 Paz, Bolivia (3.78 km). Lower UV indices at some equatorial or high-altitude sites (e.g., Quito, Ecuador) are 29 moderated by the presence of persistent cloud effects. High UVI levels (UVI > 6) are also found at most 30 mid-latitude sites during the summer months. High levels of UVI are known to lead to health problems 31 (skin cancer and eye cataracts) with extended unprotected exposures as shown in the extensive health 32 statistics maintained by Australian Institute of Health and Welfare and the United States National Institute 33 of Health National Cancer Institute.
34
35
36
37
2 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License.
38 1 Introduction
39 Calculated or measured amounts of UV radiation reaching the Earth’s surface can be used as a 40 proxy to estimate the effects of changing ozone and cloud cover on human health. High levels of UV 41 irradiance are known to affect the incidence of skin cancer (Findlay 1928, Diffey, 1987, Strom and 42 Yamamura, 1997) and the development of eye cataracts (Ambach and Blumthaler, 1993; Abraham et al., 43 2010; Roberts, 2011, Australian Institute of Health and Welfare, 2016, Howlander et al., 2019). The UV 44 response function (action spectra) for the development of skin cancer and eye cataracts are different 45 (Herman, 2010), but the effects are highly correlated. Similar correlations exists for other action spectra 46 (e.g., plant growth, vitamin D production, and DNA damage action spectra) involving the UV portion of 47 the solar spectrum. Because of these correlations, this paper will only estimate erythemal (skin reddening) 48 effects. To obtain standardized results, we use a weighted UV spectrum based on the CIE-action 49 (Commission Internationale de l'Eclairage) spectrum suggested by McKinlay and Diffey (1987) to 50 estimate the erythemal effect of UV radiation incident on human skin. erythemal irradiance (E) is usually 51 measured or calculated in energy units (mW/m2) reaching the Earth’s surface after passing through 52 atmospheric absorbing and scattering effects from ozone, aerosols, and clouds for the wavelength range 53 300 – 400 nm. A fast algorithm (Herman, 2010) for calculating E and other action spectra was developed 54 based on calculations using the scalar TUV radiative transfer program (Madronich , 1993a; 1993b; 55 Madronich and Flocke, 1997). The fast algorithm was extended to include the effect of increasing E with 56 altitude (Herman et al., 2018) as applied to the synoptic measured amounts of ozone, clouds and aerosols 57 obtained by the EPIC (Earth Polychromatic Imaging Camera) instrument onboard the DSCOVR (Deep Space 58 Climate Observatory) spacecraft orbiting about the Earth-Sun Lagrange-1 gravitational balance point 59 (Herman et al., 2018; Marshak et al., 2018).
60 This paper presents calculated noontime E(,z,t) time series and least squares LS linear trends 61 from 2005 – 2018 for 105 globally distributed locations at different latitudes , longitudes , altitudes z, 62 and time t (in years) most of which are centered on heavily populated areas such as New York City, Seoul 63 Korea, Buenos Aires, etc. (see Tables 1, 2, and 3 and an extended table A4 for 105 land sites in the 64 Appendix) based on measurements of the relevant parameters from OMI (Ozone Monitoring Instrument) 65 onboard the AURA spacecraft. An additional 77 locations in the Atlantic and Pacific Ocean for a range of 66 latitudes are also discussed. OMI satellite measurements were selected because OMI has the longest 67 continuous well calibrated UV irradiance time series from a single instrument with global coverage and 68 moderate spatial resolution (13x24 km2 at its nadir view). The derived numerical algorithms used for the 69 calculations are given in the Appendix. Estimates are given of 14-year latitude dependent changes in 70 atmospheric transmission T(,z,t) (mostly from change in cloud reflectivity), changes on total column
71 ozone TCO3(,z,tO), and changes in erythemal irradiance E(,z,tO). To augment the specific locations
72 selected for time series analysis, synoptic sunrise to sunset estimates of E(,z,tO) are derived from EPIC
73 measurements of the illuminated Earth at several Greenwich Mean Times tO with a spatial resolution of 74 18x18 km2 at the spacecraft nadir view for various longitudes centered on the Americas, Europe-Africa, 75 and Asia.
76
3 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License.
77
78 2 Erythemal Time Series and LS Linear Trends
79 Total column ozone amounts TCO3 and 340 nm Lambert Equivalent Reflectivity LER (converted to 80 transmission T(,z,t)) are retrieved from spectrally resolved irradiance measurements (300 – 550 nm) 81 obtained from OMI for the entire Earth. OMI data are filtered to remove measurements obtained from 82 portions of the CCD detector affected by the “row anomaly” (Schenkeveld et al., 2017). OMI is a polar 83 orbiting side viewing satellite instrument (2600 km width on the surface) onboard the AURA spacecraft 84 that provides near global coverage (nadir resolution field of view 13 km x 24 km) once per day from a 90- 85 minute polar orbit with an equator crossing time of approximately 13:30 local solar time (LST) (Levelt et 86 al., 2018). Because of OMI’s simultaneous side-viewing capability, there are occasionally a 2nd or 3rd data 87 points (±90 minutes) from adjacent orbits at higher latitude locations. This study uses column ozone 88 amounts (in Dobson units DU, 1 DU = 2.687x1016 molecules/cm2) and 340 nm Lambert equivalent 89 reflectivity LER (Herman et al., 2009) (LER is in reflectivity units, 0 100 Least squares linear LS trends for E(,z,t) computed from the original OMI TCO3(,z,t) and 101 LER(,t) time series having non-uniform temporal sampling give incorrect trends or slopes S(given 102 in percent change per year. Instead, LS linear trends for 105 sites (Appendix Table A4) are computed 103 from uniform temporal density (UTD method) time series based on interpolation using 2.5 times the 104 point count of the original time series. Further increases in interpolated point count N do not change the 105 S(significantly. However, the interpolated time series results yield an incorrect estimated standard 106 deviation, since ( decreases as N-0.5. Better standard deviations ( are computed from the 107 original non-uniform times series, which represents the scatter caused by the OMI non-uniform 108 sampling, intrinsic measurement noise, and atmospheric variation. Error bars shown in the various 109 graphs are statistical and do not represent possible small systematic calibration drifts in determining 110 TCO3 from the OMI instrument data (see section 3.5). 111 A standard multivariate (MV) method (Guttman, 1982) was used to check the UTD method for 112 both trends S( and trend uncertainties (. The results of the MV method (appendix Table A5) 113 comparison are based on analyzing two time series, E(t) and O3(t), for LS linear trends at each site and 114 using daily means as a reference value to estimate percent change per year. The results show that the 115 UTD method and the MV method approximately agree for S( and (for both E(t) and 4 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 116 TO3(t). Table A5 illustrates comparisons of S( and (from 5 sites showing that either method 117 may be used with comparable results. All subsequent calculations use the UTD method. 118 Erythemal Irradiance LS linear trends were also estimated from time series where the annual 119 seasonal solar zenith angle dependence is removed. The LS linear trend results were almost the same, but 120 the estimated deseasonalized error was about half the original error estimate. The original fitting error 121 estimates (are used. 122 2.1 Northern Hemisphere 123 Figure 1 and Table 1 show erythemal irradiance E(,z,t) time series (mW/m2) and their LS 124 linear trends (in percent change per year along with their 1 standard deviation) at six sites with various 125 altitudes z within the United States from 2005 – 2018 (14 annual cycles). The right-side axis shows the 126 proportional values of the standard UV index, UVI = E/25 mW/m2. Erythemal time series are truncated 127 to start and stop at the same point in their 14-year annual cycles (1 January 2014 to 31 December 2018). 128 The time series depicted in Fig.1 are non-uniform in time with significant gaps between some adjacent 129 points. In all cases the gaps are small enough to properly represent the SZA dependence of the 130 erythemal irradiance. Of the six United States sites listed in Table 1, rural Georgia (also Atlanta, GA), 131 Tampa, FL, and Honolulu HI have 2significant trends of 0.3%/Year, -0.24%/Year, and -0.27%/Year 132 (Table 1). These sites have small changes in ozone amount but significant changes in cloud + haze 133 transmission, 0.15, -0.25, and -0.24 %/Yr, respectively. Of human health interest are the maximum 134 values that occur during the summer months when the solar zenith angle is near a local minimum 135 reducing the slant column ozone absorption and Rayleigh scattering for clear-sky days. In terms of the 136 UV index, a value of 6 will produce significant skin reddening in light skinned people in about an hour of 137 unprotected exposure (Diffey, 1987; 2018; Italia and Rehfuess, 2012). In local shade, there is reduced 138 but significant exposure from atmospheric scattering (Herman et al., 1999) with the shorter more 139 damaging wavelengths scattering the most. For sites with extremely high UVI (10 – 18) even shaded 140 areas can produce significant exposure from scattered UV. Table 1 shows the 14-year average maximum 141 UVI and the 14-year average UVI. 142 For the mid-latitude site, Greenbelt, Maryland 39ON, summer values between 8 and 9 are 143 frequently reached with a few days reaching 10 and 1 day reaching 11 on 6 June 2008. The cause was a 144 low ozone value of 283 DU on a clear-sky day compared to more normal values between 310 and 340 145 DU. The basic annual cycle follows the solar zenith angle (SZA) with the minimum angle occurring during 146 the summer solstice. For Greenbelt, MD, this angle is approximately 39 – 23.3 = 15.7O. Sites with fewer 147 clouds plus haze and closer to the equator have higher maximum UV-index values, 12 for White Sands, 148 NM and 11 for Tampa, Florida. Results corresponding to Fig. 1A are summarized in Table 1 (see also 149 appendix Table A4). The last 2 columns give the estimated slope of a linear LS fit (UTD method) to each 150 time series and the standard deviation (). Graphs summarizing the 105-site Table A4 are given in 151 section 3.5, which show the expected change in UVI for decreasing latitude. Since the purpose is 152 estimating changes in E from all causes, the effects of the quasi-biennial oscillation (QBO) and solar cycle 153 are not removed from the ozone time series. 5 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 154 Fig. 1A Erythemal Irradiance E(,z,t) at six selected sites from Table A1 distributed within the United States. The red line is the linear fit to each of the time series. Also listed are the 14-year UVI average maximum and average values (UVI = E/25) (See table 1). 155 Fig. 1B Two sites from Fig. 1A, Greenbelt, Maryland and Rural Georgia, with the effect of clouds removed (i.e., T=1) 6 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Table 1 Locations in the United States (Errors are 1 Location Lat Lon Alt UVI UVI ------Trends (%/Year) ------ Deg Deg km Avg Max ERY ±Error Ozone ±Error Trans ±Error Albuquerque, NM 35.1 -106.6 1.58 6 12 0.18 0.11 -0.11 0.02 0.33 0.05 Greenbelt, MD 39 -76.9 0.06 4 10 -0.19 0.15 -0.13 0.03 -0.1 0.08 *Honolulu, HI 21.3 -157.8 0.01 9 12 -0.27 0.06 -0.02 0.01 -0.24 0.03 *Rural GA 34.5 -83.5 0.2 5 10 0.3 0.12 -0.08 0.02 0.15 0.07 *Tampa, FL 28 -82.5 0.01 7 11 -0.24 0.09 -0.03 0.02 -0.25 0.06 White Sands, NM 32.4 -106.5 1.22 7 12 -0.07 0.1 -0.07 0.02 -0.06 0.04 156 *Means 2 trend significance for erythemal change 157 When (t) = SZA and T(t) = transmission are held constant for calculating E((t), O3(t), 158 T(t)), a 1% change in total column ozone amount C produces approximately a 1.2% 159 change in erythemal irradiance. The exact amount of change is dependent on the SZA selected 160 (Eqn. 2). The values for O3 and T change (%/Year) are given in Table 1, which shows that a 161 significant amount of the erythemal irradiance change over 14 years is caused by changes in 162 cloud cover. 163 A numerical solution of the radiative transfer equation for the erythemal action 164 spectrum can be approximated by the functional form in Eq. 1 (see the appendix Eqn. A4), 165 where the cloud + scattering aerosol transmission T = (1-LER)/(1-RG), (1 > T > 0), = SZA, and RG 166 = reflectivity of the surface (an average of about 0.05) in the absence of snow or ice. This form 167 gives an improved version of the Radiation Amplification Factor R() that is independent of 168 TCO3 (Herman, 2010). 169 E() = U() (R() T (1) 170 For a given time, t, the sensitivity to changes in and dE d dT dU() dR() (2) R() R()ln E T U() 200 R() 171 Where dU() = U(d) – U() and dR() = R(d) – R() 172 For = 0, R(0) = 1.2 and R() gradually decreases to 0.85 for = 80O (Herman, 2010 and 173 Appendix Fig. A1). SZA variation is the primary anti-correlated driver for the annual cycle of 174 erythemal irradiance (Fig. 1) at each location except when there is heavy cloud cover. The cycle 175 for E(,z,t) is perturbed by the smaller effect of short-term changes in ozone amount and 176 reflectivity that are shifted in phase from (t). The result is that the separately estimated 14- 177 year linear trends for T and may not be simply additive. For example, for the Rural Georgia 178 site Fig. 1, the erythemal trend is statistically significant at 0.3 ± 0.12 %/Yr. Contributing factors 7 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 179 are the cloud transmission function T(t) trend 0.15 ± 0.07 %/Year, and small (t) trend ˗0.08 ± 180 0.02 %/Year. In Fig. 1B, the E(,z,t) trend for Rural Georgia without clouds is 0.005 ± 0.3 %/Yr. 181 The linearly combined trend is 0.15+0.08 = 0.23 ± 0.07 %/Yr, which overlaps the erythemal 182 trend error estimate. 183 For Greenbelt, the erythemal LS linear change in E(t) is -0.19 ± 0.15 %/Yr, while the 184 ozone change is ─0.13 ± 0.03 %/Yr and transmission change is -0.1 ± 0.08 %/Yr. Within 185 overlapping error estimates, these changes are consistent. The trend for the no cloud case (Fig. 186 1B) is 0.03 ± 0.3 %/Yr, but the error estimate is large enough to include the ozone change of - 187 0.13%/Yr. If the figures for Greenbelt with (Fig. 1A) and without clouds (Fig. 1B) are compared, 188 the effect of cloud cover is seen in the reduction of the maximum and mean UVI values and in 189 the strong reduction during the winter months when cloud cover is frequent. The rural Georgia 190 time series also shows a winter cloud effect that is smaller than for Greenbelt. As expected, 191 White Sands and Albuquerque, New Mexico show little winter cloud effects. 192 Figure 2A shows the latitudinal distribution, 0O to 80ON, of erythemal irradiance 193 estimated from the synoptic EPIC measurements on a line of longitude passing through San 194 Francisco, CA at 19:37 GMT or 11:37 PST. The main driver of the decrease in E() from the 195 equator toward the poles is the increased optical path from increasing SZA() and increasing O O O 196 TCO3() absorption. The smaller structure near 10 , 21 , 37 N is caused by small amounts of 197 cloud cover reducing the transmission T(). This day, 30 June 2017, near the 22 June solstice 198 was selected based on the data from DSCOVR-EPIC showing that there were few clouds present 199 in the scene (Fig.2B) with T near 1. All the E() estimates in Fig. 2A are at or near sea level and 200 yield a maximum UVI = 12 near 13ON latitude. Similarly, Fig. 2C shows the latitudinal 201 distribution of E() for the line of longitude passing near Greenwich England at 0.25OE. E() is 202 reduced because of cloud cover starting at 40ON in addition to the increasing ozone absorption 203 at higher latitudes. The accompanying images in Figs. 2B and 2D show the distribution of E() 204 and the location of significant cloud cover. 205 206 207 8 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 2A Latitudinal distribution of E() and its Fig. 2B Global distribution of E() from contributing factors, TC(O3), T, and SZA for a line of DSCOVR EPIC data on 30 June 2017 19:17 GMT longitude passing through San Francisco, CA. when there were few clouds. Fig. 2C Latitudinal distribution of E(,z,t) and its Fig. 2D Global distribution of E() from contributing factors, TC(O3), T, and SZA for a line of DSCOVR EPIC data on 04 July 2017 12:08 GMT. longitude passing near Greenwich England 208 209 2.2 Equatorial Region 210 Four selected equatorial sites (Fig. 3 and Table 2) show very different behavior 211 compared to mid-latitude sites shown in Fig. 1 and listed in Table 1. The average E(z) is 212 higher (UVI=9) for the near sea level site in Manaus, Brazil than the populated city of Quito, 213 Ecuador at 2.9 km altitude (mean UVI= 6). The lower average Quito value is caused by the 214 presence of additional cloud cover (mean transmission 9 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 215 transmission Figure 3. Four sites located close to the equator. Mt Kenya at 0.1OS, Quito Ecuador 0.2ON, Makassar Indonesia 5.1OS, Manaus Brazil 3.1ON. The blue lines are a Lowess(0.04) fit (approximately 6 month LS running average). Lowess(f) is Locally Weighted least squares fit to a fraction f of the data points, (Cleveland, 1981). 217 Figure 4A shows the effect of altitude causing an increase in clear-sky E(,z,t) for Quito 218 (2.9 km) compared to Manuas (0.1 km) plus a small difference in average TCO3 (2%) between 219 the two locations. Without clouds, both sites show a double peak corresponding to SZA = 0O 220 twice a year near the March and September equinoxes. Figure 4B has an expanded time scale 221 for 2005 showing the double peak for Quito and the strong effect of clouds in the region. The 222 average cloud-free value for Quito has a UVI = 15 and a maximum UVI = 19. The minimum 223 cloud-free value is UVI = 13 instead of 3 when cloud cover is included. The cloud effect is less at 224 inland sites at Manaus Brazil and Mt Kenya, and even at the coastal Makassar, Indonesia site. 225 The 20 DU variation in TCO3 causes the autumn peak in E(,z,t) without clouds to be smaller 226 than the spring peak. 10 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 4 Panel A: A two week running average of cloud-free E(,z,t) corresponding to the data in Fig. 3A for Quito Ecuador and Manaus Brazil showing the effect of height and a small difference in average ozone amount. Panel B: An temporal expansion for one year (2005) of E(,z,t) estimates for Quito showing the double peak as a function of minimum SZA near the equinoxes in the absence of clouds that is masked when clouds are included. The blue line shows the 20 DU variation in ozone between March and September. 227 Table 2 Summary of Equatorial Sites (Errors are 1 Location Lat Lon Alt UVI UVI ------Trends (%/Year) ------ Deg Deg km Avg Max ERY ±Error Ozone ±Error Trans ±Error *Mt Kenya, KE 0.13 37.3 5.2 14 18 -0.2 0.03 0.17 0.01 0.03 0.03 Quito, EC 0.18 -78.5 2.85 7 12 0.05 0.05 0.17 0.01 0.28 0.05 *Makassar, ID -5.13 119.4 0.01 10 14 -0.27 0.06 0.19 0.01 -0.05 0.06 Manaus, BR 3.12 -60 0.09 11 14 -0.01 0.04 0.16 0.01 0.24 0.04 228 229 The calculations based on 1O x 1O (100x100 km2) spatial resolution can obscure an 230 important health related result. In Quito, there are frequent localized clear periods when the 231 UV index can rise to the clear-sky values (13 11 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 242 transmission (Fig. 4B) is slightly decreasing a rate of -0.05 ± 0.06%/Year causing E(,z,t) to 243 have a net decrease. When combined (Fig. 4), the net effect is dominated by the increase in 244 ozone. In the absence of clouds, the percent decrease in ozone amount causes an increase in 245 E(,z,t) at approximately a 1.2:1 ratio. Figure 4B shows the approximate anti-correlation 246 between ozone amounts and E(,z,t) for Quito and Manaus. This is modified by the six-month 247 shifting of the sub-solar point (SZA = 0). When all four periodic and quasi-periodic effects are 248 combined, the result is the aperiodic function shown in Fig.4B for Quito, Ecuador. Similar 249 analysis applies for Manaus, Brazil located near the Amazon River, which is dominated by 250 variable cloud driven atmospheric transmission, but less than for Quito, Ecuador. The other two 251 equatorial sites Makassar, Indonesia and Mt. Kenya, Kenya have smaller cloud effects and show 252 periodic structures driven by SZA and ozone absorption. 253 2.3 Southern Hemisphere 254 Time series for the Southern Hemisphere are represented by six sites shown in Fig. 5 255 ranging in latitude and altitude (12.5O to 54.8O and 0 to 2.5 km). All the sites have a clear annual 256 cycle compared with the Northern Hemisphere sites. The maxima occur in January and minima 257 in June. Of these, Darwin Australia is within the equatorial zone (12.5OS) and shows the double 258 peak structure with peaks separated by about 85 days. The site furthest from the equator, 259 Ushuaia (54.8OS) has the lowest UVI peak value of 9.6 (14-year average maximum UVI = 8) and 260 a lowest 14-year minimum average UVI=2. Occasionally the Antarctic ozone depletion region 261 passes over Ushuaia giving rise to increased UV amounts, but these episodes (September – 262 October) do not correspond to the maximum UVI values that occur with the minimum SZA in 263 January. For the sites in Fig. 5, the populated site San Pedro de Atacama has the largest UVI 264 maximum (18) and average (11), since it is at moderate altitude (2.5 km) and is located at the 265 southern edge of the equatorial zone (23O) with a relatively clear cloud-free atmosphere. More 266 than half of the days each year have 10 < UVI < 18. This maximum UVI is higher than for 267 equatorial Darwin Australia, UVI < 15.5. The frequent June minima for Darwin are UVI=8 with 268 occasional days at UVI=2 caused by clouds, while the almost cloud-free San Pedro de Atacama 269 has minima of UVI=4 corresponding to a June noon SZA = 46O compared to Darwin June SZA = O 270 36 . Both sites have about the same typical TCO3, 255 DU. 271 272 Previous estimations of erythemal irradiance from measurements (1997-1999) and 273 calculations (using Total Ozone Mapping Spectrometer data) at Ushuaia (Cede et al., 2002; 274 2004) shows very similar values with a UVI < 1 in the winter (June) and with values up to 8 with 275 an occasional point reaching 10 during the summer (January) and for Buenos Aires with values 276 of UVI from 1-2 in the winter and up to 12-13 in the summer. These values approximately 277 agree with those in Table 3. The Cede et al. (2004) results for 8 sites also include a higher 278 altitude equatorial site, La Quiaca, AR (22.1OS), at 3.46 km altitude, having summer values up to 12 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 279 UVI = 20. The corresponding calculated estimates using OMI data (2005-2018) also have the 280 maximum UVI = 20 occurring in 2010 with a 14-year average maximum of UVI=18 (Table 3). La 281 Quiaca has decreasing E(t) caused by increasing cloudiness and increasing TCO3. 282 Fig. 5: Six sites in the Southern Hemisphere including estimates of the trends for E(,z,t), TC(O3), and the atmospheric transmission T caused by clouds and haze. The TC(O3) time series (blue) is shown for Ushuaia 283 284 285 286 287 288 289 290 291 292 293 294 13 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Table 3 Summary for 7 Southern Hemisphere Sites (Errors are 1 Location Lat Lon Alt UVI UVI ------Trends (%/Year) ------ Deg Deg km Avg Max ERY ±Error Ozone ±Error Trans ±Error Darwin, AU -12.5 120.8 0.01 11 14 0.04 0.05 0.12 0.01 0.15 0.04 La Quiaca, AR -22.1 -65.6 3.46 12 18 -0.15 0.07 0.09 0.01 -0.12 0.04 San Pedro CL -22.9 -68.2 2.45 11 17 -0.0 0.08 0.15 0.01 0.06 0.02 Queenstown, SA -31.9 26.92 1.1 7 14 -0.1 0.11 0.16 0.01 0.01 0.05 Buenos Aires, AR -34.6 -58.4 0.03 6 13 -0.2 0.14 0.08 0.02 -0.1 0.07 Melbourne, AU -37.3 145 0.01 5 13 -0.4 0.15 0.25 0.02 -0.24 0.06 Ushuaia AR -54.8 -68.3 0.06 2 8 0.01 0.2 0.18 0.03 0.05 0.07 295 Fig. 6A Latitudinal distribution of E(,z,t) and its Fig. 6B Global distribution of E() from contributing factors, TC(O3), T, and SZA for a line of DSCOVR EPIC data on 31 December 2017 longitude passing near Sydney, Australia 02:24:36 GMT. 296 297 Figure 6A shows the latitudinal distribution, 0O to 80OS, of erythemal irradiance on a line 298 of longitude passing near Sydney Australia at 02:24:36 GMT or 1:24 NSW (New South Wales). 299 The main driver of the decrease in E() from the equator toward the poles is the increased O 300 optical path from increasing SZA() and the increasing TCO3(). The smaller structures near 40 , 301 50O, 60OS are caused by small amounts of cloud cover reducing the transmission T(). The 31 302 December day near the solstice was selected based on data from DSCOVR-EPIC showing that 303 there were few clouds present over Australia (Fig.6B). All the E() estimates in Fig. 6A are near 304 sea level with a maximum UVI = 14 near 25OS latitude. Figure 6B shows the distribution of high 305 E() over Australia and Indonesia with the highest values in Australia for 31 December 2017. 14 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 306 The differences between the northernmost city (Darwin) and the southernmost city 307 (Melbourne) are quite large in terms of UV exposure because of differences in SZA, ozone 308 amount, and cloud cover leading to a larger number of days per year with high UVI, a few 309 weeks for Melbourne and three months for Darwin. This is reflected in the non-melanoma skin 310 cancer statistics published by Australian Institute of Health and Welfare (2016) for the different 311 regions with the Northern Territories (containing Darwin) having double the rate per 100,000 312 people compared to Victoria containing Melbourne (Pollack et al., 2014). Fig. 7 Monthly average variation in E(,z,t) for six sites in both the Northern and southern Hemispheres. Solid lines are from data summarized in a World Health Organization study. https://www.who.int/uv/intersunprogramme/activities/uv_index/en/index3.html. The small numbers are the height on the histogram bars (W/m2). 313 The values of E(,z,t) over Antarctica, Fig. 6, are likely not accurate because the 314 reflectivity of the scene is approximately treated as if there were a thin cloud over a bright 315 surface. The calculated transmission function T(t) has a minimum of 0.89 resulting in a 316 difference in E(,z,t) between setting T = 1 and using the Antarctic Peninsula calculated T(=- 15 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 317 70, =-64) of less than 10%. The annual cycle ranges from 0 in winter (May to August) to a 318 variable maximum in December depending on the year. For example, 125 mW/m2 in 2013 and 319 175 mW/m2 in2016. The year to year variation in the maximum E(,z,t) is driven the variable 320 ozone amount. The largest amount at =-70O, =-64O occurred in 2013 and the smallest amount 321 was in 2016 as measured by OMI. 322 The monthly averages in E(,z,t) shown in Fig. 7 have the expected strong variation 323 with SZA and with latitude in both the Northern and Southern Hemispheres. Of the sites 324 shown, the smallest values are in Ushuaia, Argentina with a peak average value in February of 325 108 W/m2 (UVI = 4) and a minimum in June of 4 W/m2. However, there are days when both 326 ozone values are below 200 DU and the SZA is near its minimum (about 32O) giving rise to UVI 327 values of 8 (Fig. 5). This contrasts with the Los Angeles California site in Fig. 7 where the 328 monthly average maximum is 254 W/m2 (UVI = 10) and the average over 14 years of the daily 329 maximum is 11 (see Table A4). Figure 7 shows a comparison of monthly average E(,z,t) for 3 330 sites with a World Health Organization compilation of E(,z,t) for the 21st of each month (solid 331 line). 332 3 Global View of E(,tO) distributions from DSCOVR EPIC 333 EPIC onboard the DSCOVR spacecraft views the sunlit disk of the Earth from a small orbit 334 about the Earth-Sun gravitational balance point (Lagrange-1 or L1) 1.5 million kilometers from 335 the Earth. EPIC has 10 narrow band filters ranging from the UV at 310 nm to the near infrared, 336 870 nm that enable measurements of TCO3 and LER with 18 km nadir resolution using a 2048 337 x2048 pixel charge coupled detector. EPIC takes multiple (12 to 22) sets of 10 wavelength 338 images per day as the Earth rotates on its axis. The instrumental details and calibration 339 coefficients for EPIC are given in Herman et al. (2018) as well as some examples of UV 340 estimates. 341 EPIC measured UV irradiances are derived from measured TCO3 and 388 nm LER for 342 about 3 million grid points as shown in for 22 June 2017 at 06:13 GMT (Fig. 8). These quantities 343 along with terrain height maps are converted into E(,z,tO) for each grid point at the specified 344 GMT time tO using the algorithm given in the appendix. R = LER is converted into transmission T 345 using T = (1 – R)/(1 – RG), where RG is the surface reflectivity (Herman and Celarier, 1997), on 346 average RG is approximately 0.05, for most scenes without snow or ice. The simple expression 347 for T gives approximately the same results as a more elaborate analysis of clouds and aerosols 348 averaged over large scenes as seen from EPIC (Krotkov et al., 2001). The LER map in Fig. 8 can 349 be compared to the color image of the Earth obtained by EPIC, where the high values of LER 350 correspond to the bright clouds shown in the color image. Ozone absorption mostly affects the 351 short wavelength portion of the erythemal spectrum (300 – 320 nm), with only small 16 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 352 absorption from 340 – 400 nm. The effects of Rayleigh scattering are also included. The results 353 of combining TCO3, LER and Rayleigh scattering to estimate erythemal irradiance are shown in 354 Fig. 9 (upper left) for 22 June 2017 with tO = 06:13 GMT. Fig. 8 EPIC derived TCO3 (upper left in DU: 100 to 500 DU) and reflectivity (LER upper right in percent or RU: 0 to 100) for 22 June 2017 at tO = 06:13 GMT. Lower left: color image of the Earth showing clouds and land areas. The brighter clouds are optically thick and correspond to the higher values of the LER. 355 The data from EPIC are synoptic (same GMT) so that the ozone, reflectivity, and 356 erythemal results are from sunrise (west or left) to sunset (right or east) with decreasing values 357 for SZA near sunrise and sunset. A similar erythemal darkening effect from increased SZA occurs 358 for north and south higher latitudes. In these images, local solar noon is near the center, but 359 offset by EPIC’s viewing angle that is 4O to 15O away from the Earth-sun line. In the case shown, 360 the six-month orbit is offset about 10O to the west. Three months earlier in March and three 361 months later in September, the orbit is offset to the east. 362 Erythemal maps in subsequent figures are organized by season (December and June 363 solstices, and March and September equinoxes). The maximum values of E(,tO) follow the 17 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 364 minimum SZA modified by cloud amount. Since the sub-solar point moves with the annual 365 change in the Earth’s declination angle (between ±23.3O), the maximum UVI usually occurs at 366 local solar noon (LST) with the smallest SZA. An exception is when the effect of increased 367 altitude is larger than the SZA effect. An example of this is shown in Fig. 10 for the Himalayan 368 Mountains, which contain Mt Everest. The maximum UVI for Mt Everest in Fig. 8 is 18 even 369 though it is at about 10:30 LST (SZA = 7O) 370 3.1 Northern Hemisphere Summer Solstice (June) 371 For the June solstice view (Fig.9), the EPIC view includes the entire Arctic region and 372 areas to about 55OS. The center line of the image is close local solar noon. The view is with 373 north up and from sunrise (west or left) to sunset (east or right). The effect of the orbital 374 distance of 4O to 15O from the Earth-Sun line can be seen in the asymmetry of the sunrise and 375 sunset regions implying that that the six-month orbit was off to the south-west of the Earth-sun 376 line. The images were selected to give estimates of erythemal irradiance over Asia, Africa, and 377 the Americas as a function of latitude, altitude, and longitude (time of the day) for a specific 378 Greenwich Mean Time (GMT) for each map. 379 In Fig. 9 high UVI is seen over the Himalayans Mountains (06:13 GMT) and central 380 western China, reaching over UVI = 18 for a June day on Mt Everest (28.0ON, 86.9OE, 8.85 km). 381 On the next view in Fig. 9 over central Africa (11:21 GMT) there are elevated UVI =11, and on 382 the third image in Fig. 9 (19:00 GMT) there is an elevated UVI area over the mountainous 383 regions of Mexico (UVI = 16). The effect of significant cloud cover at moderate SZA can be seen 384 (blue color), where the UVI is reduced to 2 near noon (e.g., Gulf of Mexico at 19:00 GMT). There 385 are reductions in E(,z,t) from lower reflectivity clouds in the center of the Fig.9 images that 386 are not easily seen in the UVI image with the expanded scale (0 to 20). The effect of higher 387 reflectivity clouds in Fig. 8 are easily seen in Fig. 9 in blue color representing low amounts of 388 E(,z,t) at the ground. There are only small percent change features in the ozone distribution, 389 so that few ozone related structures are expected in the E(,tO) images for such a coarse UVI 390 scale. 18 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig.9 Erythemal irradiance E(,z,t) and UVI from sunrise to sunset for 21 June 2017 solstice. The three images are for different GMT. Upper left 22 June 2017 (06:22GMT). Upper Right 21 June 2017 (11:21 GMT) and Lower Left 21 June 2017 (19:00 GMT). The images correspond to the sub-solar points over different continents caused by the Earth’s rotation (15O per hour). 391 19 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 10A E(,tO) near the summit of Mt. Everest at Fig. 10B Erythemal irradiance on 22 June 2017 an altitude of 8.85 km. Mean T=0.7 near Mt. Everest within the Tibetan Plateau region (red color) in mW/m2. 392 O 393 The E(,tO) time series for Mt Everest (Fig. 10A) at 8.85 km altitude and at 28 N has a 394 mean value of UVI = 10 and has an average annual peak occurring in June with UVI = 18 on days 395 with little or no cloud cover. For the same conditions, except for artificially setting the altitude 396 at sea level, the maximum UVI = 13. The sea level mean UVI value is 7. There is an average net 397 altitude correction for maximum UVI of (18 - 13)/(13*8.8) = 4.3%/km and 5.6%/km for the 398 mean UVI, including corrections for ozone amount and latitude (Appendix Eqn. A7). O O 399 Figure 10B shows the distribution of E(,tO) around Mt Everest (approximately 2 x 2 ) 400 for 22 June 2017 at 06:13 GMT (Fig. 9) and the effect of heavy cloud cover (blue and purple 401 areas). The reflectivity in Fig. 8, shows there is a mixture of light and heavy cloud cover in the 402 region reducing the amount of UV reaching the surface. The red region in the upper left is the 403 location of the Mt Everest peak with very high values of UVI. The yellow colors represent high 404 UVI values of about 10 and green about 8. Winter values at the top of Everest ae quite low as 405 shown in Fig. 10A ranging from UVI =2 to 4 depending on cloud cover. 406 407 408 409 410 20 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 411 3.2 E(,tO) for September and March Equinox Conditions Fig. 11A E(,tO) and UVI from sunrise to sunset for 21 September 2017 equinox. The three images are for different GMT 412 413 Near the September and March equinoxes (Figs. 11A and 11B) the sun is overhead near 414 the Equator giving high UVI = 12 in many areas with higher values (16 to 18) in the mountain 415 regions (e.g., Southern Indonesia, Peru’s Andes Mountains, and some high-altitude regions in 416 Malawi and Tanzania. While the Sun-Earth geometry is nearly the same for both equinoxes, 417 there is considerable difference in seasonal cloud cover for the two equinox days. The area of 418 sub-Saharan Africa near Nigeria has particularly high UVI values caused by nearly cloud-free 419 conditions over a wide region implying a considerable heath risk for mid-day UV exposure. 420 Other high UVI values occur over smaller elevated areas. This is particularly evident in the 421 nearly cloud-free high-altitude Peruvian Andes at about 28OS even though the SZA = 28O. 21 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 11B E and UVI from sunrise to sunset for 21 March 2017 equinox. The three images are for different GMT (05;33, 10:56, and 16:20). 422 423 3.3 Southern Hemisphere Summer Solstice (December) 424 During the December solstice the sun is overhead at 23.3OS (Fig. 12). The reduced SZA 425 causes high UVI levels throughout the region between 20OS and 40OS especially in elevated 426 regions such as the Chilean Andes, Western Australia and elevated regions of southeastern 427 Africa (South Africa, Tanzania, Kenya). For the case where Western Australia is near local solar 428 noon, the UVI levels reach about 13 to 14 between 20OS to 34OS, a region that includes the city 429 of Perth with more than a million people and several smaller cities and towns. These high UVI 430 values represent a considerable health risk for skin cancer, since the UVI stays above 12 for 431 nearly a month. The same is true for eastern Australia (Fig. 6) during December that implies the 432 high skin cancer risk for the entire Australian continent. The same comments apply to New 433 Zealand, eastern South Africa and elevated areas further north (e.g., Tanzania Africa). Even 22 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 434 higher values occur in the Andes Mountains in Chile and Peru that include some small cities 435 (see Fig. 4 for San Pedro de Atacama, Peru). Fig. 12 E(tO) and UVI from sunrise to sunset for 21 December 2017 solstice. The three images are for different GMT. 436 3.4 Erythemal Synoptic Variation (Sunrise to Sunset) 437 The longitudinal dependence of E(tO) is illustrated in Fig. 13 where sunrise to sunset slices O O 438 have been taken for an equatorial latitude, 0.1 N and mid-latitude, 30.85 N. The estimated E(tO) 439 includes the effect of clouds and haze (Panels C and D) included in the atmospheric transmission 440 function T(tO) and the effects of local terrain height. The maximum E(tO) is to the east of the sub- 441 satellite point because the satellite orbit about the Lagrange-1 point L1 is displaced to the west of the 442 Earth-Sun line on 14 April 2016. The northward displacement is caused by the Earth’s declination angle 443 of about 9.6O. This corresponds to the minimum SZA shown in Fig. 13A Panel A of 9.5O. Panels A, B, C, D 444 show the effects of cloud transmission for all values of LER that are not easily seen in the global 445 erythemal color maps (bottom panels of Fig. 13A. The presence of clouds is easily seen in the color 446 image for 14 April at 4:21 GMT (Fig. 13B). 23 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 447 O O Fig. 13A Longitudinal slices of E(tO) (units UVI) at 0.1 N and 30.85 N latitude shown by the dark horizontal bars. The EPIC E(tO) images are for 14 April 2016 tO = 04:21 GMT centered at about O O O 10 N and 104 E. Panels A and C show longitudinal slices of E(tO) and T(tO) for = 0.1 N and panels B and D for 30.85ON. The solid lines in panels A and B represent the SZA. 448 24 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig 13B EPIC color image for 14 April 2016 at Fig 13C EPIC scene reflectivity LER for 14 April 2016 04:12:16 GMT showing the distribution of cloud at 04:12:16 GMT cover and land corresponding to Fig. 13A O O 449 The main cause of the decrease of E(tO) with latitude between 0.1 N and 30.85 N is caused 450 by the increased SZA followed by the latitudinal increase in TCO3. The difference is modulated (Panels A 451 and B) by the presence of clouds and haze (Figs. 13B and 13C) and haze in T(tO) shown in Fig. 13A 452 Panels C and D. There are nearly clear-sky patches for the equatorial sample leading to very high UVI = 453 14 compared to the mid-latitude maximum of UVI = 9 because of the effect of clouds near the time of 454 minimum SZA. The distribution of clouds is shown in the true color picture of the Earth obtained by EPIC 455 on 14 April 2016 at 04:12:16 GMT centered on 104OE. The bright white portion of cloud mages are the 456 optically thick clouds of high reflectivity and low transmission. 457 3.5 Zonal average E(tO) and 14-Year Trends 458 Figure 14 shows a summary of the zonal maximum (Panel A) and zonal average (Panel B) UVI 459 values on 14 April 2016 at 04:21 GMT from Fig. 13A for longitudinal bands plots from -75O < Latitude < 460 75O. The solid lines are a smooth Akima spline fit (Akima, 1970) to the data points. Depending on the day 461 of the year, the location of the maximum will shift between -23.45O to +23.45O following the position of 462 overhead sun. The zonal average maximum (Fig. 14A) of about UVI = 14 is approximately the same for 463 any day of the year. This includes longitudes containing high altitude sites at moderately low latitudes 464 where the local UVI maximum can reach 18 to 20. The US Environmental Protection Agency classifies 465 exposure at UVI=6 to 7 as high, which requires protection for extended exposure (e.g., 1 hour). For low 466 latitudes, UVI > 6 occurs several hours around local solar noon. For equatorial latitudes at sea level, UVI 467 > 6 occurs for about 6 hours (Fig. 13). The zonal average values (Fig.14B) are considerably smaller, since 468 they are more affected by clouds than the mostly clear-sky maxima in Fig. 14A. 25 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 14 Zonal Maximum UVI (Panel A), Zonal Average (Panel B) on 14 April 2016 at 04:21 GMT from EPIC including the effect of clouds and haze, as a function of latitude. Both the data points and an Akima spline fit are shown. 469 470 Of interest are similar analyses of the 105 land sites as listed in the Appendix Table A4 and 471 summarized in Figs. 15 and 16. The results over 14 years show much higher levels of UVI than for the 472 single day zonal average shown in Fig. 14, especially for the four indicated high-altitude sites. The 473 maximum summer values at all latitudes between 60OS and 60ON exceed UVI = 6, which is considered 474 high enough to cause sunburn for unprotected skin (Sánchez-Pérez et al., 2019) in 20 to 50 minutes 475 depending on skin type. Higher values of UVI can produce sunburn in much shorter times. For example, 476 for UVI = 10, sunburn can be produced in as little as 15 to 30 minutes exposure of unprotected skin. 477 The highest UVI values in Table A4 are associated with 4 sites at high altitudes. Two of these are 478 populated cities, San Pedro de Atacama (Population = 11,000), Chile and La Paz Bolivia (Population = 479 790,000). These two sites have very high UVI because of their altitude, low latitude, and relative lack of 480 clouds on some days. Over the 14 years of this study, the UVI at San Pedro de Atacama has remained 481 approximately constant while at La Paz, Bolivia the UVI has decreased at a rate of 4.6 ± 0.05% per 482 decade caused by an increase in ozone amount (1.4 ± 0.1 % per decade) and a decrease in atmospheric 483 transmission (-3.2 ± 0.5 % per decade). 484 26 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 15. Fourteen-year UVI Average and UVI Maximum from Table A4 for 105 sites. Solid curves are Akima spline fits to the individual site data points. There are 4 high altitude sites listed, San Pedro, Chile (2.45 km), La Paz, Bolivia (3.78 km), Mt Kenya, Kenya (5.2 km), and Mt Everest, Nepal and China (8.85 km). 485 Table A4 also presents the 14-year linear LS trends of changes in erythemal irradiance, 486 Atmospheric Transmission, and column ozone amount (%/Yr) and the 1- error estimate for those 487 trends. The results are summarized in Figs.16 for the 105 selected land sites. There is significant 488 variation in atmospheric transmission, ± 0.2%/Yr, (mostly cloud reflectivity) for an extended latitude 489 range, 55OS – 35ON. However, on average there is no systematic change as indicated by the local least 490 squares fit (red line Lowess(0.5)). For latitudes greater than 40ON atmospheric transmission has 491 increased (cloud reflectivity decreased) for the period 2005 to 2018 implying that solar insolation has 492 also increased for all UV (305 – 400 nm), visible wavelengths (400 – 700 nm), and near infrared 493 wavelengths (700 – 2000 nm). For the UV portion of the spectrum represented by the erythemal 494 irradiance action spectrum, the change is affected by changes in TCO3. The TCO3 changes (Fig. 16C) 495 result in an average decrease in irradiance for latitudes between 55OS and 35ON and a smaller %/Yr 496 increase for higher latitudes than would be expected based on non-absorbing atmospheric transmission 497 changes. The ozone changes obtained from OMI observations include the effects of the 11.3-year solar 498 cycle, the quasi-biennial oscillation QBO, and the El Nino Southern Oscillation ENSO effects, and, as such, 499 are not the standard ozone trend amounts (Weber et al., 2017). 27 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 16 Least squares (LS) percent change per year for (A) Erythemal Irradiance, (B) Atmospheric Transmission, and (C) Column Ozone for the period 2005 – 2018 from OMI observations at 105 individual sites (see Table A4). The solar cycle and quasi- biennial oscillation effects have not been removed. Error bars are 1. Red curve is a Lowess(0.3) fit to the data. 500 501 Similar changes can be estimated over the Atlantic and Pacific oceans for an extended latitude 502 range from 60OS to 60ON without intersecting land at 30OW (Atlantic) and 179OW (Pacific) in steps of 5O 503 latitude (Fig. 17). The percent change per year over the Atlantic Ocean at 30OW at high northern 504 latitudes is the opposite of those occurring over land with a decrease in transmission (increase in 505 reflectivity) implying a decrease in solar insolation. A similar analysis over the Pacific Ocean at 179OW 506 shows a change that shows an increase in transmission at high northern latitudes of the same 507 magnitude as occurs over land implying increased solar insolation over a wide wavelength range (380 – 508 2000 nm). In the UV range the erythemal irradiance changes follows the changes in transmission offset 509 by the smaller changes in column ozone amount. 510 The band of equatorial cloud reflectivity has decreased (transmission increased) for both the 511 Atlantic and Pacific Oceans at 0O and at 5ON. For the Pacific at 179OW, the estimated changes correspond 512 to the El Nino Southern Oscillation ENSO region suggesting a decrease in cloud reflectivity. On either 513 side at 5ON and 10OS there is a decrease in transmission of about 2% per decade over the Atlantic and 514 about 4 to 5%/decade over the Pacific ENSO longitude. 515 28 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 17: Similar to Fig. 16 but over the Atlantic (Longitude 32OW) and the Pacific (179OW) Oceans with one data point every 10O of latitude. 516 517 Figure 18 shows the changes that have occurred over the major landmasses as a function of latitude 518 over specified longitudes, A. Europe Africa 20OE, B. North America 90OW, B. South America 60OW, and D. 519 Russia-China-India 120OW. The results are quite variable with North America showing the increase in the 520 rate of transmission T increase at high latitudes offset in E by a small increase in TCO3. Europe-Africa also 521 shows the increase in the rate of T increase that is bigger in effect than a small decrease in TCO3. South 522 America shows little change, but the northern part of the graph is over the Atlantic Ocean at 60OW and 523 shows rates that are different from North America at 90OW. 524 29 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 18 Similar to Fig. 17 but for land areas as indicated for longitudes 20OE, 90OW, 60OE, 120OW 30 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 525 Fig. 19 Zonal average of change in OMI column ozone amount and in atmospheric transmission (%/Yr). The red line in Fig. 19B is a Lowess(0.5) fit showing the general trend as a function of latitude. 526 O 527 Figure 19A shows the zonal average percent change per year computed every 2.5 for TCO3( 528 and for the atmospheric transmission function T(TCO3 is showing increases for Southern latitudes 529 and for latitudes up to 20ON. As mentioned earlier, the ozone trend includes the effects of the solar 530 cycle, QBO, and ENSO effects, which is appropriate for this study of erythemal irradiance and its 531 changes. The NASA OMI project suggests that there may be an OMI drift of +0.1% per year (private 532 communication) relative to a reference TCO3 data set derived from the overlap (2012 – 2018) with 533 NOAA 19 SBUV/2 (National Oceanographic Atmospheric Administration Solar Backscatter Ultra Violet - 534 2) instrument. The effect of this systematic drift would be to shift the curve in Fig. 19A downward by 535 0.1%/Year or be considered as an uncertainty that is greater than the small statistical uncertainties. 536 Figure 19B shows the zonal average percent change per year for atmospheric transmission T() caused 537 by the presence of aerosols and clouds where T(t) has been normalized to the assumed invariant 538 Antarctic high plateau ice reflectivity. The results indicate that there is on average increased solar 539 insolation for high northern latitudes. The decreased cloud cover suggests a positive feedback 540 mechanism for global warming. 541 4 Summary 542 Measured total column ozone TCO3 and Lambert Equivalent Reflectivity LER (converted to 543 atmospheric transmission) data T(,z,t) from AURA-OMI have been combined along with terrain height 544 data to estimate noon time series for erythemal irradiance E(,z,t) in mW/m2 (or UVI = E/25) reaching 545 the Earth’s surface at globally distributed specified locations using Eqns. A1 to A9. This paper 546 summarizes the results from 182 land plus ocean locations, some having dangerously high values of UVI 547 caused either by the presence of low SZA and ozone values or high altitudes under almost clear-sky 548 conditions. For some sites, there has been no long-term LS linear change (2005 – 2018) in UVI at the 549 two-standard deviation level 2. However, nearly half the sites have shown 2 changes in UVI caused by 550 changes in atmospheric transmission (clouds plus aerosols) and an offset from zero caused by changes in 31 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 551 ozone amount. Fourteen-year atmospheric transmission trends are calculated showing little change in 552 average T (mostly cloud reflectivity) from 55OS to 35ON, but with significant increase in T from 40ON to 553 60ON causing increased solar insolation from the UV to NIR wavelengths at these latitudes suggesting 554 positive feedback from global warming. TCO3 also shows significant latitudinal change with an increase 555 between 55OS to 35ON and a decrease from 40ON to 60ON that only affects UV wavelengths (300 – 340 556 nm). The maximum UVI is shown for each selected site with, as expected, low latitudes and elevated 557 sites showing the highest UVI values (14 to 18) compared to typical NH mid-latitude sites at low altitude 558 having a maximum UVI = 8 to 10. The OMI based results show agreement with monthly average values 559 data summarized in a World Health Organization study and with measurements of UVI made in 560 Argentina (Cede et al., 2002; 2004). Global synoptic maps of UVI from sunrise to sunset are shown from 561 DSCOVR/EPIC data for specific days corresponding the solstices and equinoxes. These show the high UVI 562 values occurring at local solar noon over wide areas and especially at high altitudes and the decrease 563 with SZA caused by latitude and solar time. Figure 14 shows a zonal average for 14 April 2016 from EPIC 564 data showing latitudes of very high UVI that track the seasonal solar declination angle corresponding to 565 hemispheric summer. Similarly, Fig.15 shows the zonal average of the 105 land sites in Table A4 that 566 includes 4 very high-altitude sites with UVI = 18. The EPIC and OMI observations show that there are are 567 the wide areas between 20O and 30OS latitude during the summer solstice in Australia (Fig.12) showing 568 near noon values with UVI = 14, values that are dangerous for production of skin cancer and eye 569 cataracts and correlate with Australian National Institute of Health and Welfare cancer incidence health 570 statistics (2016). Similar values of high UVI occur for the latitude range ±30O that includes parts of Africa 571 and Asia. Two equatorial region high altitude cities, San Pedro, Chile (2.45 km), La Paz, Bolivia (3.78 km), 572 with frequently clear sky conditions have very high UVIMAX = 17 and 18 and UVIAVG = 11 in contrast to 573 Quito, Ecuador (2.85 km) that has substantial cloud cover UVIMAX = 11 and UVIAVG = 7. Cities located at 574 sea level in the equatorial zone also can have high vales of UVIMAX = 15 (e.g., Lima, Peru). 575 576 32 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 577 Appendix 578 Some of the contents of this appendix are reproduced for convenience from Herman et al. (2018) 579 and Herman (2010). Fitting error estimates from solutions of the radiative transfer equations are given in 580 Herman (2010). The notation used in Herman (2010) and Herman et al., 2018 is retained with SZA = Solar 581 Zenith Angle, = SZA, = total column ozone amount in DU TCO3, = wavelength in nm, and CT = 582 fractional cloud + haze transmission T. An improved numerical fit for the altitude dependence is provided 583 for Eqn. A7 and for the coefficients in Eqn. A8. 2 584 Erythemal irradiance E0(CT) at the Earth’s sea level (W/m ) is defined in terms of a 585 wavelength dependent weighted integral over a specified weighting function A() times the incident solar 2 586 irradiance I(ΩCT) (W/m ) (Eq. A1). The erythemal weighting function Log10(AERY()) is given by the 587 standard erythemal fitting function shown in Eq. A2 (McKinley and Diffey, 1987). Tables of radiative O 588 transfer solutions for DE = 1 AU are generated for a range of SZA (0 < < 90 ), for ozone amounts 100 < 589 600 DU, and terrain heights 0 < Z < 5 km using an approximation to the solutions from the TUV DISORT 590 radiative transfer model as described in Herman (2010) for erythemal and other action spectra (e.g., plant 591 growth PLA, vitamin-D production VIT, cataracts CAT, etc.). The irradiance weighted by the erythemal 592 action spectrum is given by 400 E ( ,,C ) I (, ,,C ) A()d 0 T T 250 (A1) 250 < < 298 nm Log10(AERY) = 0 (A2) 298 < < 328 nm Log10(AERY) = 0.094 (298 - ) 328 < < 400 nm Log10(AERY) = 0.015 (139 - ) 593 Equation A1 can be closely approximated by the power law form (Eq. A3), where U() and R() are 594 fitting coefficients (R()) is an improved Radiation Amplification Factor that is independent of ) to the 595 radiative transfer solutions in the form of rational fractions (Herman, 2010). Rational fractions were 596 chosen because they tend to behave better at the ends of the fitting range than polynomials with 597 comparable fitting accuracy. -R() (A3) E0(,CT) = U() () CT U() or R() = (a+c2+ex4)/(1+b2+d4+f6) r2 > 0.9999 (A4) (A5) CT = (1-LER)/(1-RG) where RG is the reflectivity of the surface 2 (A6) E(z E0() H(,z) / DE 33 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Numerically, H(,z) is calculated by forming the ratio RE = E(zE0() Where most of the and dependence is contained in EO (A7) H(,Z) = [(-3.8443E-3 Zkm+3.1127E-4) /200 + 0.054111 Zkm+1] G( ) (A8) G() = g+h+i2+j3+k4 The coefficients a, b, c, d, e, f, g, h, j, and k are in Tables A1 and A2 598 599 When Eq. A6 is applied to the ozone and LER data, the global E(z at the Earths’ surface can be 600 obtained after correction for the Earth-Sun distance DE, where DE in AU can be approximated by (Eq. A9), 601 DE = 1 − 0.01672 cos(2π (day_of_year − 4)/365.25) (A9) 602 Table A1 Coefficients R() and coefficient U() for 0 < < 80O Eq. A4 and 100 < < 600 DU for E() = U() (R() ( 1.0E10 = 1.0x1010) U() or R() = (a+c2+e4)/(1+b2+d4+f6) r2 > 0.9999 (see Fig. A1) Action Spectra U() (watts/m2) R() CIE Erythemal a= 0.4703918683355716 a= 1.203020609002682 b= 0.0001485533527344676 b= -0.0001035585455444773 UERY & RERY c= -0.0001188976502179551 c= -0.00013250509260352 d= 1.915618238117361E-08 d= 4.953161533805639E-09 e= 7.693069873238405E-09 e= 1.897253186594168E-09 f= 1.633190561844982E-12 f= 0.0 Table A2 Solar Zenith angle function G() used in Eq. A8 G() = g+h+i2+j3+k4 g= 9.999596516311959E-01 j= 1.752907417831904E-07 h= 2.384464204972423E-05 k= -2.482705952292921E-09 i= 3.078822311353050E-06 603 604 Since RE() has only weak and dependence an approximation can be obtained by forming the 605 mean of RE over and Then a linear approximation is 606 H(z) = 1 + 0.047 Zkm (A10) 607 Equation A10 is similar to Eqn. A7 with G()=1 and = 300 DU 34 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 608 H(300, Z) = 1 + 0.052 Zkm (A12) 609 Table A4 Summarizes the erythemal irradiance E(z and its rate of change for specific 610 locations (Latitude, Longitude, and Altitude) based on the algorithm from Eqns. A1 - A9. CT includes the 611 effect of both cloud and aerosol transmission to the surface (for non-absorbing aerosols). Absorbing 612 aerosols (ephemeral smoke and dust aerosols) are not included. Also included are the rates of change for 613 ozone over the 14-year period. 614 Sites that have trends statistically significant at the two-standard deviation level (96% probability) 615 for E(zare indicated with an *. For a number of sites, E(z can show significant change even when 616 there is almost no change in where the change in E(z is caused by increases or decreases in CT. 617 The expected change in EO(z with ozone change ranges from about 0.82 to 1.2 (see R() in Table A1 618 and Fig. A1) depending on the latitude (SZA as a function of latitude). Sites deviating significantly from this 619 ratio have been affected by changes in cloud transmission. 620 Fig. A1 Values of the coefficients RERY() and UERY() 621 622 623 624 35 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 625 A similar approximate analysis can be obtained for height dependence of other action spectra given by 626 Herman (2010) for Zkm = 0 and the references therein. Table A3 Height Dependence of Six Action Spectra Action Spectrum Approximate Height Dependence Vitamin-D VIT 1 + 0.055 Zkm Cataracts CAT 1 + 0.050 Zkm DNA Damage DNA 1 + 0.056 Zkm Erythemal ERY 1 + 0.047 Zkm Plant damage PLC 1 + 0.046 Zkm Plant Damage PLA 1 + 0.038 Zkm 627 Height dependence increases for those action spectra with more emphasis on shorter UV wavelengths 628 629 Table A4 lists 105 city or land locations in various countries as indicated in alphabetical order. Table A4 Erythemal UV Index and Linear Change for UVI, O3, and Transmission 2005 – 2018 for 105 locations Location Lat Lon Alt UVI UVI ------Trends (%/Year) ------ Deg Deg km Avg Max Ery ± Ozone ± Trans ± Abu Dhabi AE* -24.47 54.37 0.01 8 14 -0.17 0.08 0.16 0.01 0.11 0.04 Abuja, HG 9.07 7.49 0.01 10 13 -0.5 0.05 0.09 0.01 -0.34 0.05 Accra GH 5.56 -0.19 0.03 10 13 -0.05 0.05 0.14 0.01 0.12 0.04 Adelaide, AU* -34.92 138.6 0 6 13 -0.45 0.14 0.17 0.02 -0.18 0.05 Albuquerque, NM 35.1 -106.6 1.58 6 12 0.18 0.11 -0.11 0.02 0.33 0.05 Algiers DZ 36.75 3.04 0.19 5 10 0.23 0.13 -0.08 0.02 0.37 0.05 Alice Springs, AU* -23.7 133.88 0.58 9 15 -0.29 0.08 0.11 0.01 -0.12 0.04 Anchorage, AK 61.1 -149.9 0.03 2 6 0.37 0.2 -0.33 0.03 0.63 0.08 Athens, GR 37.98 23.73 0.72 5 11 0.02 0.14 -0.07 0.02 0.26 0.05 Atlanta, GA 33.5 -84.5 0.31 6 11 -0.2 0.12 -0.1 0.02 -0.41 0.07 Auckland, NZ -36.85 174.76 0.05 5 13 0.02 0.14 0.22 0.02 0.16 0.06 Bangalore, IN* 12.97 77.58 0.91 10 14 -0.21 0.05 0.14 0.01 0.03 0.05 Bangkok TH* 13.74 100.52 0.01 9 13 -0.25 0.05 0.14 0.01 -0.01 0.04 Bangor, ME 44.81 -68.8 0.05 3 9 -0.17 0.17 -0.13 0.03 -0.08 0.08 Beijing, CN 39.9 116.4 0.05 4 9 0.25 0.13 -0.1 0.02 0.04 0.06 Bogota CO* 4.62 -74.06 2.54 10 15 -0.73 0.05 0.21 0.01 -0.43 0.05 Boston, MA* 42.36 -71.05 0.04 4 9 0.76 0.16 -0.13 0.03 0.62 0.08 Brasilia, BR* -15.83 -47.93 1.17 10 15 0.16 0.06 0.12 0.01 0.22 0.05 Brisbane AU -27.47 153.03 0.03 8 14 -0.18 0.1 0.14 0.02 0.01 0.05 Buenos Aires, AR -34.6 -58.4 0.03 6 13 -0.15 0.14 0.08 0.02 -0.1 0.07 Bulawayo, ZW -20.15 28.58 1.35 9 15 0.13 0.07 0.13 0.01 0.23 0.05 Canberra, AU* -35.28 149.13 0.57 6 13 -0.41 0.13 0.17 0.02 -0.27 0.06 Cape Town ZA* -33.92 18.42 0.04 7 13 -0.24 0.12 0.19 0.02 -0.02 0.04 Caracas, VZ* 10.5 -66.92 0.9 11 14 -0.09 0.04 0.13 0.01 0.12 0.03 Casablanca MA 33.59 -7.6 0.15 6 11 0 0.11 -0.07 0.02 0.11 0.04 36 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Chennai, IN* 13.07 80.24 0.02 10 13 -0.41 0.05 0.12 0.01 -0.19 0.05 Chicago, US* 41.89 -87.67 0.19 4 10 -0.42 0.17 -0.04 0.03 -0.37 0.09 Christchurch, NZ -43.53 172.64 0.02 4 11 -0.11 0.17 0.33 0.03 -0.1 0.07 Cordoba AR -31.41 64.18 0.4 7 13 -0.15 0.12 0.21 0.02 0.12 0.05 Darwin, AU* -12.5 120.8 0.01 11 14 0.04 0.05 0.12 0.01 0.15 0.04 Des Moines, IA 41.6 -93.6 0.29 4 10 0.01 0.17 -0.06 0.02 0.55 0.09 Detroit, MI 42.3 -83.05 0.19 4 9 0.21 0.17 -0.08 0.03 0.4 0.09 Dhaka, BD* 23.7 90.4 0.01 8 12 -0.47 0.07 0.14 0.01 -0.19 0.06 Dongguan CN 23.02 113.75 0.01 6 12 -0.45 0.09 0.06 0.02 -0.37 0.09 Eureka, CA* 40.8 -124.1 0.01 4 10 0.33 0.14 -0.08 0.03 0.4 0.06 Giza EG 30 31.21 0.03 7 11 -0.06 0.09 0.06 0.02 0.11 0.03 Greenbelt, MD 39 -76.9 0.06 4 10 -0.19 0.15 -0.13 0.03 -0.1 0.08 Hamilton, NZ* -37.85 175.32 0.05 5 12 -0.31 0.15 0.25 0.03 -0.13 0.07 Hartford, CT 41.8 -72.8 0.01 4 9 0.12 0.16 -0.12 0.03 0.1 0.08 Helsinki, FI 61.92 25.75 0.01 2 6 -0.22 0.21 -0.06 0.03 -0.07 0.09 Honolulu, HI* 21.3 -157.8 0.01 9 12 -0.27 0.06 -0.02 0.01 -0.24 0.03 Iowa Center, IA 42 -93.5 0.3 4 10 0 0.17 -0.02 0.03 0.58 0.1 Iquitos, PE* -3.75 -73.25 0.13 9 14 -0.41 0.06 0.14 0.01 -0.23 0.06 Jakarta, ID* -6.21 106.85 0.08 9 13 -0.27 0.05 0.18 0.01 -0.04 0.06 Kinshasa, CD -4.32 15.31 0.31 10 14 -0.08 0.05 0.17 0.01 0.13 0.05 La Quaca, AR* -22.11 -65.57 4.46 12 18 -0.15 0.07 0.09 0.01 -0.12 0.04 Lagos, NG* 6.47 3.41 0.01 9 13 -0.44 0.07 0.1 0.01 -0.25 0.07 Lauder, NZ* -45.05 169.7 0.37 4 11 0.43 0.18 0.31 0.03 0.26 0.07 La Paz, BO* -16.5 -68.15 3.78 11 18 -0.46 0.05 0.14 0.01 -0.32 0.05 Leeds, UK* 53.8 -1.55 0.03 2 7 -0.43 0.2 0.09 0.03 -0.2 0.08 Lima PE* -12.04 -77.03 0.15 10 15 -0.35 0.08 0.17 0.01 -0.2 0.05 London, UK 51.51 -0.12 0.02 2 7 0.12 0.19 0.04 0.03 0.01 0.08 Los Angeles, CA 34.5 -118.5 0.1 6 11 -0.06 0.11 -0.11 0.02 0.09 0.04 Madrid, ES 40.42 -3.7 0.65 5 10 -0.27 0.14 -0.07 0.02 -0.24 0.06 Makassar, ID* -5.13 119.4 0.01 10 14 -0.27 0.06 0.19 0.01 -0.05 0.06 Manaus, BR 3.12 -60 0.09 11 14 -0.01 0.04 0.16 0.01 0.24 0.04 Manhattan, NY 40.76 -73.97 0.01 4 9 0.02 0.15 -0.19 0.03 -0.26 0.08 Marin County, CA 37.5 -122 0.1 6 11 0.04 0.12 -0.1 0.02 0.12 0.05 Mauna Loa Obs., HI* 19.54 155.6 3.4 11 15 -0.17 0.06 0.11 0.02 0.01 0.04 Melbourne, AU* -37.3 145 0.01 5 13 -0.4 0.15 0.25 0.02 -0.24 0.06 Mendoza, AR* -32.9 -68.9 0.83 7 14 -0.33 0.12 0.13 0.02 -0.12 0.04 Mexico City, MX* 19.43 -99.13 2.24 9 14 -0.26 0.06 0.07 0.01 -0.11 0.05 Moscow, RU* 55.75 37.62 0.15 2 7 0.47 0.22 -0.04 0.03 0.61 0.1 Mt_Everest_0 28 86.9 0 7 13 0 0.08 0.08 0.01 0.3 0.06 Mt_Everest_8.85 28 86.9 8.85 10 18 0 0.08 0.08 0.01 0.3 0.06 Mt. Kenya, KE* 0.13 37.3 5.2 14 18 -0.2 0.03 0.17 0.01 0.03 0.03 Mumbai, IN* 19.08 72.88 0.02 9 13 -0.14 0.06 0.12 0.01 0.04 0.05 Nairobi, KE* 1.09 35.88 1.86 12 15 -0.25 0.04 0.16 0.01 -0.04 0.03 New Delhi, IN 28.61 77.2 0.03 7 12 -0.06 0.08 0.11 0.01 0.09 0.04 Nice FR 43.67 7.29 0.03 4 9 0.01 0.15 -0.13 0.03 -0.05 0.06 37 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Palembang, ID* -2.99 104.76 0.01 9 14 -0.27 0.06 0.2 0.01 0 0.06 Paris, FR* 48.86 2.35 0.04 3 8 0.47 0.18 -0.01 0.03 0.51 0.08 Perth, AU -31.95 115.9 0.03 7 14 -0.03 0.12 0.21 0.02 0.22 0.04 Pilar, AR -31.66 -63.88 0.34 7 14 -0.02 0.12 0.07 0.02 -0.09 0.06 Punta Arenas, CL -53.16 -70.92 0.04 3 9 -0.15 0.2 0.17 0.02 -0.15 0.07 Queenstown, SÁ -31.9 26.92 1.1 7 14 -0.12 0.11 0.16 0.01 0.01 0.05 Quezon City PH 14.65 121.05 0.05 8 13 -0.12 0.07 0.2 0.01 0.21 0.06 Quito, EC 0.18 -78.5 2.85 7 12 0.05 0.05 0.17 0.01 0.28 0.05 Reading, CA 40.5 -122.4 0.03 5 10 0.2 0.14 -0.02 0.03 0.3 0.06 Recife, BR -8.05 -34.93 0.55 11 14 -0.08 0.05 0.14 0.01 0.13 0.03 Rio de Janeiro BR -22.9 -43.21 0 8 14 0.19 0.1 0.17 0.01 0.15 0.07 Riyadh, SÁ* 24.77 46.68 0.62 9 13 -0.17 0.07 0.09 0.01 0.03 0.03 Rome, IT 41.9 12.5 0.01 4 10 0.13 0.15 -0.16 0.02 0.3 0.06 Rosario, AR* -32.94 -60.64 0.03 6 14 -0.3 0.13 0.08 0.02 -0.26 0.06 Rural Georgia, GA* 34.5 -83.5 0.2 5 10 0.3 0.12 -0.08 0.02 0.15 0.07 Sacramento, CA 38.5 -121.5 0.08 5 10 0 0.13 -0.09 0.02 0.15 0.05 Salt Lake, City UT 40.7 -111.9 1.32 5 11 0.24 0.16 -0.08 0.02 0.48 0.07 San Julian, AR -49.32 -67.75 0.06 3 10 0.14 0.18 0.16 0.02 0.13 0.06 Santa Rosa, CA 38.5 -122.7 0.05 5 10 0.12 0.13 -0.09 0.02 0.19 0.05 Santiago CL -33.46 -70.65 0.56 7 14 -0.06 0.15 0.17 0.02 0.39 0.08 San Jose, CA 37.5 -122.5 0.14 5 10 0.23 0.12 -0.1 0.02 0.31 0.05 San Pedro, CL -22.9 -68.2 2.45 11 17 -0.04 0.08 0.15 0.01 0.06 0.02 Seattle, WA 47.5 -123.5 0.14 3 9 0.3 0.19 -0.09 0.03 0.29 0.09 Shanghai, CN* 31.22 121.47 0.06 5 11 -0.32 0.13 -0.02 0.02 -0.27 0.09 Stanley FK -51.7 -57.9 0.05 3 9 0.18 0.19 0.12 0.02 0.15 0.07 Tampa, FL* 28 -82.5 0.01 7 11 -0.24 0.09 -0.03 0.02 -0.25 0.06 Tel Aviv, IL 32.11 34.86 0.03 7 11 -0.12 0.11 -0.01 0.02 -0.01 0.03 Tokyo JP* 35.65 139.84 0.04 4 10 0.42 0.13 0.05 0.02 0.52 0.08 Ushuaia, AR -54.8 -68.3 0.06 2 8 0.01 0.2 0.18 0.03 0.05 0.07 Utah Center, UT 39 -109.5 1.8 5 11 -0.15 0.14 -0.1 0.02 0 0.06 Vientiane, LA 17.97 102.63 0.17 9 12 -0.03 0.05 0.11 0.01 0.12 0.05 Wellington NZ -41.3 174.8 0.08 5 12 -0.6 0.16 0.31 0.03 -0.35 0.06 White Sands, NM 32.4 -106.5 1.22 7 12 -0.07 0.1 -0.07 0.02 -0.06 0.04 630 631 E is the slope S ± of the linear least squares fit to the time series E(t) with 1 standard deviation 632 633 The same notation applies to the ozone time series O3(t) and transmission T(t). 634 Two independent methods, Uniform Temporal Distribution UTD and Multivariate MV (Guttman, 635 1982), were used to calculate LS linear trends and their uncertainties ±σ showing that the methods 636 yielded similar results (Table A5) 637 638 38 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Table A5 Comparison of UTD and MV methods of trend and uncertainty estimation Location Method E(t) Trend ±σ TO3(t) Trend ±σ (%/yr) (%/yr) (%/yr) (%/yr) Greenbelt, MD MV -0.22 0.22 -0.14 0.04 (φ = 39.0o) UTD -0.19 0.15 -0.13 0.03 Rural Georgia, GA MV 0.27 0.19 -0.09 0.03 (φ = 34.5o) UTD 0.30 0.12 -0.08 0.02 Buenos Aires, AR MV -0.15 0.21 0.07 0.03 (φ = -34.6o) UTD -0.20 0.14 0.08 0.02 Melbourne, AU MV -0.36 0.23 0.25 0.04 (φ = -37.5o) UTD -0.40 0.15 0.25 0.02 Ushuaia, AR MV 0.02 0.31 0.18 0.04 (φ = -54.8o) UTD 0.01 0.20 0.18 0.03 639 640 641 642 643 644 645 646 647 648 649 650 651 39 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 652 5.0 References 653 Abraham, Alison G., Christopher Cox, Sheila West, The Differential Effect of Ultraviolet Light Exposure on 654 Cataract Rate across Regions of the Lens, Investigative Ophthalmology & Visual Science August, Vol.51, 655 3919-3923. doi:10.1167/iovs.09-4557, 2010. 656 Akima, Hiroshi, A new method of interpolation and smooth curve fitting based on local procedures, J. 657 ACM, 17(4), 589-602, 1970 658 Ambach, W. and Blumthaler, M., Biological effectiveness of solar UV radiation in humans, Experientia, 659 49: 747. https://doi.org/10.1007/BF01923543, 1993. 660 Australian Institute of Health and Welfare, Skin cancer in Australia. Cat. no. CAN 96. Canberra: AIHW, 661 2016. 662 Cede, Alexander, Luccini, Eduardo, Núñez, Liliana, Piacentini, Rubén, Blumthaler, Mario, Monitoring of 663 erythemal irradiance in the Argentine ultraviolet network. Journal of Geophysical Research. 107. 664 10.1029/2001JD001206, 2002. 665 Cede, Alexander, Luccini, Eduardo, Núñez, Liliana, Piacentini, Rubén, Blumthaler, Mario, Herman, Jay, 666 TOMS-derived erythemal irradiance versus measurements at the stations of the Argentine UV 667 Monitoring Network. Journal of Geophysical Research. 109. 10.1029/2004JD004519, 2004. 668 Cleveland, William S., LOWESS: A program for smoothing scatterplots by robust locally weighted 669 regression. The American Statistician. 35 (1): 54. JSTOR 2683591. doi:10.2307/2683591, 1981. 670 671 Diffey BL, Analysis of the risk of skin cancer from sunlight and solaria in subjects living in northern 672 Europe, Photo-dermatology, 4(3):118-126], 1987. 673 Diffey BL. Time and place as modifiers of personal UV exposure. Int J Environ Res Public Health, 674 15(6):E1112, doi: 10.3390/ijerph15061112, 2018. 675 Findlay, G. M., Ultra-Violet Light and Skin Cancer. Lancet pp.1070-73 ref.14, 1928. 676 Guttman, I., Linear Models, An Introduction, 358 pp., Wiley-Interscience, New York, 1982. 677 Krotkov, N. A., J. R. Herman, P. K. Bhartia, V. Fioletov, and Z. Ahmad, Satellite estimation of spectral 678 surface UV irradiance, 2. Effects of homogeneous clouds and snow, J. Geophys. Res., 106, 11,743– 679 11,759, 2001. 680 Herman, J.R. and E Celarier, J. Geophys. Earth surface reflectivity climatology at 340-380 nm from TOMS 681 data, 102, 28003-28011, 1997. 682 Herman, J.R., N. Krotkov, E. Celarier, D. Larko, and G. Labow, Distribution of UV radiation at the Earth’s 683 surface from TOMS-measured UV-backscattered radiances, J. Geophys. Res., 104, D10, 12,059–12,076, 684 1999. 40 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 685 Herman, J.R., G. Labow, N.C. Hsu, D. Larko, Changes in Cloud Cover (1998-2006) Derived From 686 Reflectivity Time Series Using SeaWiFS, N7-TOMS, EP-TOMS, SBUV-2, and OMI Radiance Data, J. 687 Geophys. Res., 114, D01201, doi:10.1029/2007JD009508, 2009. 688 Herman, J.R., Use of an improved radiation amplification factor to estimate the effect of total ozone 689 changes on action spectrum weighted irradiances and an instrument response function, J. Geophys. 690 Res., D23119, doi:10.1029/2010JD014317, 2010 691 Herman, J., Huang, L., McPeters, R., Ziemke, J., Cede, A., and Blank, K.: Synoptic ozone, cloud reflectivity, 692 and erythemal irradiance from sunrise to sunset for the whole Earth as viewed by the DSCOVR 693 spacecraft from the Earth–sun Lagrange 1 orbit, Atmos. Meas. Tech., 11, 177-194, 694 https://doi.org/10.5194/amt-11-177-2018, 2018. 695 Howlader N, Noone AM, Krapcho M, Miller D, Brest A, Yu M, Ruhl J, Tatalovich Z, Mariotto A, Lewis DR, 696 Chen HS, Feuer EJ, Cronin KA (eds). SEER Cancer Statistics Review, 1975-2016, National Cancer Institute. 697 Bethesda, MD, https://seer.cancer.gov/csr/1975_2016/, based on November 2018 SEER data 698 submission, posted to the SEER web site, April 2019. 699 Italia, Nadia, Eva A. Rehfuess, Is the Global Solar UV Index an effective instrument for promoting 700 sun protection? A systematic review, Health Education Research, Volume 27, Issue 2, Pages 200– 701 213, https://doi.org/10.1093/her/cyr050, 2012 702 Levelt, P. F., Joiner, J., Tamminen, J., Veefkind, J. P., Bhartia, P. K., Stein Zweers, D. C., Duncan, B. N., 703 Streets, D. G., Eskes, H., van der A, R., McLinden, C., Fioletov, V., Carn, S., de Laat, J., DeLand, M., 704 Marchenko, S., McPeters, R., Ziemke, J., Fu, D., Liu, X., Pickering, K., Apituley, A., González Abad, G., 705 Arola, A., Boersma, F., Chan Miller, C., Chance, K., de Graaf, M., Hakkarainen, J., Hassinen, S., Ialongo, I., 706 Kleipool, Q., Krotkov, N., Li, C., Lamsal, L., Newman, P., Nowlan, C., Suleiman, R., Tilstra, L. G., Torres, O., 707 Wang, H., and Wargan, K.: The Ozone Monitoring Instrument: overview of 14 years in space, Atmos. 708 Chem. Phys., 18, 5699-5745, https://doi.org/10.5194/acp-18-5699-2018, 2018. 709 Madronich, S. , The atmosphere and UV-B radiation at ground level, in Environmental UV Photobiology, 710 edited by L. O. Björn, and A. R. Young, pp. 1–39, Plenum, New York, 1993a 711 Madronich, S. , UV radiation in the natural and perturbed atmosphere, in Environmental Effects of UV 712 (Ultraviolet) Radiation, edited by M. Tevini, pp. 17–69, A. F. Lewis, Boca Raton, 1993b. 713 Madronich, S., and S. Flocke , Theoretical estimation of biologically effective UV radiation at the Earth’s 714 surface, in Solar Ultraviolet Radiation—Modeling, Measurements and Effects, NATO ASI Series,vol. I52, 715 edited by C. Zerefos, Springer, Berlin, 1997. 716 McKinlay, A., and B. L. Diffey, A reference action spectrum for ultraviolet-induced erythema in human 717 skin; in Human Exposure to Ultraviolet Radiation: Risks and Regulations, Int. Congress Ser., edited by W. 718 F. Passchier, and B. F. M. Bosnajakovic, pp. 83–87, Elsevier, Amsterdam, Netherlands, 1987. 41 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 719 Pollack, A. McGrath, M. Henderson, J. Britt, H. Skin cancer by state and territory, Australian Family 720 Physician, The Royal Australian College of General Practitioners (RACGP), 507, 721 http://www.racgp.org.au/afp/2014/august/skin-cancer-by-state-and-territory, 2014. 722 723 Roberts, Joan, Ultraviolet Radiation as a Risk Factor for Cataract and Macular Degeneration, Eye & 724 Contact Lens: Science & Clinical Practice. 37(4):246-249,DOI:10.1097/ICL.0b013e31821cbcc9, PMID: 725 21617534, 2011. 726 727 Sánchez-Pérez, J. F., Vicente-Agullo, D., Barberá, M., Castro-Rodríguez, E., and Cánovas, M., Relationship 728 between ultraviolet index (UVI) and first-, second- and third-degree sunburn using the Probit 729 methodology, Nature Scientific Reports, 9, 2045-2322, https://doi.org/10.1038/s41598-018-36850-x, 730 2019. 731 732 Schenkeveld, V. M. E. and Jaross, G. and Marchenko, S. and Haffner, D. and Kleipool, Q. L. and 733 Rozemeijer, N. C. and Veefkind, J. P. and Levelt, P. F., In-flight performance of the Ozone Monitoring 734 Instrument, Atmospheric Measurement Techniques, 10, 1957-1986, 2017. 735 736 Strom SS, Yamamura Y., Epidemiology of nonmelanoma skin cancer, Clinics in Plastic Surgery, 24(4):627- 737 636,1997 738 Weber, Mark, Coldewey-Egbers, Melanie, Fioletov, Vitali, Frith, Stacey & Wild, Jeannette & P. Burrows, 739 John & S. Long, Craig & Loyola, Diego, Total ozone trends from 1979 to 2016 derived from five merged 740 observational datasets - the emergence into ozone recovery. Atmospheric Chemistry and Physics 741 Discussions. 2017. 1-37. 10.5194/acp-2017-853, 2017. 742 42 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 743 Figure Captions 744 745 Fig, 1A Erythemal Irradiance E(,z,t) at six selected sites from Table A1 distributed within the United 746 States. The red line is the linear fit to each of the time series. Also listed are the 14-year UVI average 747 maximum and average values (UVI = E/25) (See table 1) 748 749 Fig. 1B Two sites from Fig. 1A, Greenbelt, Maryland and Rural Georgia, with the effect of clouds 750 removed (i.e., T=1) 751 752 Fig. 2A Latitudinal distribution of E() and its contributing factors, TC(O3), T, and SZA for a line of 753 longitude passing through San Francisco, CA. 754 755 Fig. 2B Global distribution of E() from DSCOVR EPIC data on 30 June 2017 19:17 GMT when there 756 were few clouds. 757 758 Fig. 2C Latitudinal distribution of E(,z,t) and its contributing factors, TC(O3), T, and SZA for a line of 759 longitude passing near Greenwich England 760 761 Fig. 2D Global distribution of E() from DSCOVR EPIC data on 04 July 2017 12:08 GMT. 762 763 Fig. 3. Four sites locate close to the equator. Mt Kenya at 0.1OS, Quito Ecuador 0.2ON, Makassar 764 Indonesia 5.1OS, Manaus Brazil 3.1ON. The blue lines are a Lowess(0.04) fit (approximately 6 month 765 running average). 766 767 Fig. 4 Panel A: A two week running average of cloud-free E(,z,t) corresponding to the data in Fig. 3A 768 for Quito Ecuador and Manaus Brazil showing the effect of height and a small difference in average 769 ozone amount. Panel B: An expansion for one year (2005) of E(,z,t) estimates for Quito showing the 770 double peak as a function of minimum SZA near the equinoxes in the absence of clouds that is masked 771 when clouds are included. The blue line shows the 20 DU variation in ozone between March and 772 September. 773 774 Fig. 5: Six sites in the Southern Hemisphere including estimates of the trends for E(,z,t), TC(O3), and 775 the atmospheric transmission T caused by clouds and haze. The TC(O3) time series (blue) is shown for 776 Ushuaia 777 778 Fig. 6A Latitudinal distribution of E(,z,t) and its contributing factors, TC(O3), T, and SZA for a line of 779 longitude passing near Sydney, Australia 780 781 Fig. 6B Global distribution of E() from DSCOVR EPIC data on 31 December 2017 02:24:36 GMT. 782 43 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 783 Fig. 7 Monthly average variation in E(,z,t) for six sites in both the Northern and southern 784 Hemispheres. Solid lines are from data summarized in a World Health Organization study. 785 https://www.who.int/uv/intersunprogramme/activities/uv_index/en/index3.html. The small 786 numbers are the height on the histogram bars (W/m2). 787 788 Fig. 8 EPIC derived ozone amount (upper left in DU: 100 to 500 DU) and reflectivity (LER upper right in 789 percent or RU: 0 to 100) for 22 June 2017 at tO = 06:13 GMT. Lower left: color image of the Earth 790 showing clouds and land areas. The brighter clouds are optically thick and correspond to the higher 791 values of the LER. 792 793 Fig.9 Erythemal irradiance E(,z,t) and UVI from sunrise to sunset for 21 June 2017 solstice. The three 794 images are for different GMT. Upper left 22 June 2017 (06:22GMT). Upper Right 21 June 2017 (11:21 795 GMT) and Lower Left 21 June 2017 (19:00 GMT). The images correspond to the sub-solar points over 796 different continents caused by the Earth’s rotation (15O per hour). 797 798 Fig. 10A E(,tO) near the summit of Mt. Everest at an altitude of 8.85 km. Mean T=0.7 799 800 Fig. 10B Erythemal irradiance on 22 June 2017 near Mt. Everest within the Tibetan Plateau region (red 801 color) in mW/m2 802 803 Fig. 11A E(,tO) and UVI from sunrise to sunset for 21 September 2017 equinox. The three images are 804 for different GMT 805 806 Fig. 11B E and UVI from sunrise to sunset for 21 March 2017 equinox. The three images are for different 807 GMT (05;33, 10:56, and 16:20). 808 809 Fig. 12 E(tO) and UVI from sunrise to sunset for 21 December 2017 solstice. The three images are for 810 different GMT 811 O O 812 Fig. 13A Longitudinal slices of E(tO) (units UVI) at 0.1 N and 30.85 N latitude shown by the dark O 813 horizontal bars. The EPIC E(tO) images are for 14 April 2016 tO = 04:21 GMT centered at about 10 N O O 814 and 104 E. Panels A and C show longitudinal slices of E(tO) and T(tO) for = 0.1 N and panels B 815 and D for 30.85ON. The solid lines in panels A and B represent the SZA. 816 817 Fig 13B EPIC color image for 14 April 2016 at 04:12:16 GMT showing the distribution of cloud cover and 818 land corresponding to Fig. 13A 819 820 Fig 13C EPIC scene reflectivity LER for 14 April 2016 at 04:12:16 GMT 821 44 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 822 Fig. 14 Zonal Maximum UVI (Panel A), Zonal Average (Panel B) on 14 April 2016 at 04:21 GMT from EPIC 823 including the effect of clouds and haze, as a function of latitude. Both the data points and an Akima 824 spline fit are shown. 825 826 Fig. 15. Fourteen-year UVI Average and UVI Maximum from Table A4 for 105 sites. Solid curves are 827 Akima spline fits to the individual site data points. There are 4 high altitude sites listed, San Pedro, Chile 828 (2.45 km), La Paz, Bolivia (3.78 km), Mt Kenya, Kenya (5.2 km), and Mt Everest, Nepal and China (8.85 829 km). 830 831 Fig. 16 Least squares (LS) percent change per year for (A) Erythemal Irradiance, (B) Atmospheric 832 Transmission, and (C) Column Ozone for the period 2005 – 2018 from OMI observations at 105 833 individual sites (see Table A4). The solar cycle and quasi-biennial oscillation effects have not been 834 removed. Error bars are 1. Red curve is a Lowess(0.3) fit to the data. 835 836 Fig. 17: Similar to Fig. 16 but over the Atlantic (Longitude 32OW) and the Pacific (179OW) Oceans with 837 one data point every 10O of latitude. 838 839 Fig. 18 Similar to Fig. 16 but for land areas as indicated for longitudes 20OE, 90OW, 60OE, 120OW 840 841 Fig. 19 Zonal average of change column ozone amount and in atmospheric transmission (%/Yr). The red 842 line in Fig. 19B is a Lowess(0.5) fit showing the general trend as a function of latitude. 843 844 45 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 845 6.0 Author Contributions 846 847 Jay Herman is responsible for all the text, figures, erythemal algorithm, and trend determinations 848 849 Liang Huang is responsible for deriving Lambert Equivalent Reflectivities for the OMI and EPIC 850 instruments and ozone for the EPIC instrument. He is also responsible for the in-flight calibration of 851 the EPIC instrument’s UV channels. 852 853 Alexander Cede and Matthew Kowalski are responsible for the stray light correction, and “flat- 854 fielding” of the EPIC CCD 855 856 Karin Blank is responsible for the ongoing improvements in geolocation and determining the correct 857 exposure times for the EPIC instrument. 858 859 Jerald Ziemke is responsible for verifying the method of linear least-squares trend determination used 860 to analyze the OMI time series data 861 862 863 Author List 864 865 Jay Herman1 Alexander Cede2 Liang Huang3 Jerald Ziemke5 866 Matthew Kowalewski2, Karin Blank4 867 1University of Maryland Baltimore County JCET, Baltimore, Maryland USA 868 [email protected] 869 870 2SciGlob Instruments and Services, Ellicott City, Maryland, USA 871 [email protected] 872 [email protected] 873 874 3Science Systems and Applications, Lanham, Maryland, USA 875 [email protected] 876 877 4NASA Goddard Space Flight Center, Greenbelt, Maryland USA 878 [email protected] 879 880 5Morgan State University, GESTAR, Baltimore Maryland 881 [email protected] 882 883 7.0 Acknowledgements The authors would like to thank and acknowlege the support of the DSCOVR 884 project and the OMI science team for the OMI satellite project for making OMI data freely available 46 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 885 8.0 Figures Fig. 1A Erythemal Irradiance E(,z,t) at six selected sites from Table A1 distributed within the United States. The red line is the linear fit to each of the time series. Also listed are the 14-year UVI average maximum and average values (UVI = E/25) (See table 1) Fig. 1B Two sites from Fig. 1A, Greenbelt, Maryland and Rural Georgia, with the effect of clouds removed (i.e., T=1) 47 Figure 1A and Figure 1B https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 886 Fig. 2A Latitudinal distribution of E() and its Fig. 2B Global distribution of E() from contributing factors, TC(O3), T, and SZA for a line of DSCOVR EPIC data on 30 June 2017 19:17 GMT longitude passing through San Francisco, CA. when there were few clouds. Fig. 2C Latitudinal distribution of E(,z,t) and its Fig. 2D Global distribution of E() from contributing factors, TC(O3), T, and SZA for a line of DSCOVR EPIC data on 04 July 2017 12:08 GMT. longitude passing near Greenwich England 887 888 Figures 2A 2B 2C 2D 48 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 889 Fig. 3. Four sites locate close to the equator. Mt Kenya at 0.1OS, Quito Ecuador 0.2ON, Makassar Indonesia 5.1OS, Manaus Brazil 3.1ON. The blue lines are a Lowess(0.04) fit (approximately 6 month running average). 890 891 Figure 3 892 49 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 4 Panel A: A two week running average of cloud-free E(,z,t) corresponding to the data in Fig. 3A for Quito Ecuador and Manaus Brazil showing the effect of height and a small difference in average ozone amount. Panel B: An expansion for one year (2005) of E(,z,t) estimates for Quito showing the double peak as a function of minimum SZA near the equinoxes in the absence of clouds that is masked when clouds are included. The blue line shows the 20 DU variation in ozone between March and September. 893 894 Figure 4 895 50 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 5: Six sites in the Southern Hemisphere including estimates of the trends for E(,z,t), TC(O3), and the atmospheric transmission T caused by clouds and haze. The TC(O3) time series (blue) is shown for Ushuaia 896 897 898 Figure 5 899 51 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 6A Latitudinal distribution of E(,z,t) and its Fig. 6B Global distribution of E() from contributing factors, TC(O3), T, and SZA for a line of DSCOVR EPIC data on 31 December 2017 longitude passing near Sydney, Australia 02:24:36 GMT. 900 901 902 Figure 6A 6B 903 52 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 7 Monthly average variation in E(,z,t) for six sites in both the Northern and southern Hemispheres. Solid lines are from data summarized in a World Health Organization study. https://www.who.int/uv/intersunprogramme/activities/uv_index/en/index3.html. The small numbers are the height on the histogram bars (W/m2). 904 905 906 907 Figure 7 908 53 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 909 910 911 912 913 914 Fig. 8 EPIC derived ozone amount (upper left in DU: 100 to 500 DU) and reflectivity915 (LER upper right in percent or RU: 0 to 100) for 22 June 2017 at tO = 06:13 GMT. Lower left: color image of the Earth 916 showing clouds and land areas. The brighter clouds are optically thick and 917 correspond to the higher values of the LER. 918 919 920 921 922 Figure 8 923 54 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig.9 Erythemal irradiance E(,z,t) and UVI from sunrise to sunset for 21 June 2017 solstice. The three images are for different GMT. Upper left 22 June 2017 (06:22GMT). Upper Right 21 June 2017 (11:21 GMT) and Lower Left 21 June 2017 (19:00 GMT). The images correspond to the sub-solar points over different continents caused by the Earth’s rotation (15O per hour). 924 925 926 927 928 Figure 9 929 55 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 10A E(,tO) near the summit of Mt. Everest Fig. 10B Erythemal irradiance on 22 June 2017 at an altitude of 8.85 km. Mean T=0.7 near Mt. Everest within the Tibetan Plateau region (red color) in mW/m2 930 931 932 933 934 935 936 Figures 10A 10B 937 56 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 11A E(,tO) and UVI from sunrise to sunset for 21 September 2017 equinox. The three images are for different GMT 938 939 940 941 942 943 Figure 11A 57 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 944 Fig. 11B E and UVI from sunrise to sunset for 21 March 2017 equinox. The three images are for different GMT (05;33, 10:56, and 16:20). 945 946 947 948 949 Figure 11B 58 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. 950 Fig. 12 E(tO) and UVI from sunrise to sunset for 21 December 2017 solstice. The three images are for different GMT. 951 952 953 Figure 12 954 59 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. O O Fig. 13A Longitudinal slices of E(tO) (units UVI) at 0.1 N and 30.85 N latitude shown by the dark horizontal bars. The EPIC E(tO) images are for 14 April 2016 tO = 04:21 GMT centered at about O O O 10 N and 104 E. Panels A and C show longitudinal slices of E(tO) and T(tO) for = 0.1 N and panels B and D for 30.85ON. The solid lines in panels A and B represent the SZA. 955 Figure 13A 60 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig 13B EPIC color image for 14 April 2016 at Fig 13C EPIC scene reflectivity LER for 14 April 2016 04:12:16 GMT showing the distribution of cloud at 04:12:16 GMT cover and land corresponding to Fig. 13A 956 957 958 959 960 961 962 963 Figure 13B 964 61 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 14 Zonal Maximum UVI (Panel A), Zonal Average (Panel B) on 14 April 2016 at 04:21 GMT from EPIC including the effect of clouds and haze, as a function of latitude. Both the data points and an Akima spline fit (Akima, 1970) are shown. 965 966 967 968 969 970 971 Figure 14 972 62 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 15. Fourteen-year UVI Average and UVI Maximum from Table A4 for 105 sites. Solid curves are Akima spline fits (Akima, 1970) to the individual site data points. There are 4 high altitude sites listed, San Pedro, Chile (2.45 km), La Paz, Bolivia (3.78 km), Mt Kenya, Kenya (5.2 km), and Mt Everest, Nepal and China (8.85 km). 973 974 975 976 977 978 979 Figure 15 980 63 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 16 Least squares (LS) percent change per year for (A) Erythemal irradiance, (B) Atmospheric Transmission, and (C) Column Ozone for the period 2005 – 2018 from OMI observations at 105 individual sites (see Table A4). The solar cycle and quasi- biennial oscillation effects have not been removed. Error bars are 1. Red curve is a Lowess(0.3) fit to the data. 981 982 983 984 985 986 987 Figure 16 988 64 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 17: Similar to Fig. 16 but over the Atlantic (Longitude 32OW) and the Pacific (179OW) Oceans with one data point every 10O of latitude. 989 990 991 992 993 Figure 17 994 65 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 18 Similar to Fig. 16 but for land areas as indicated for longitudes 20OE, 90OW, 60OE, 120OW Figure 18 66 https://doi.org/10.5194/acp-2019-793 Preprint. Discussion started: 21 October 2019 c Author(s) 2019. CC BY 4.0 License. Fig. 19 Zonal average of change column ozone amount and in atmospheric transmission (%/Yr). The red line in Fig. 19B is a Lowess(0.5) fit showing the general trend as a function of latitude. 995 996 997 998 999 1000 1001 1002 Figure 19 1003 67